COMPUTATIONAL TOOLS TO ENABLE THE DESIGN OF SMART SOFT MATERIALS

(POTENTIAL) 

''ERDF A way of making Europe''

(01/09/2023 – 31/08/2026)

The discovery, design, and manufacturing of new soft smart materials such as Electro Active Polymers (EAPs) and Magneto Active Polymers (MAPs), has been reported to be essential for Europes productivity and competitiveness, for businesses of all sizes and across all sectors of the economy. If the EU is to become a true geopolitical actor, it cannot shy away from this challenge. In this vein, the SpanisH Science, Technology, and Innovation Strategy (EECTI) 2021-2027 identifies the discovery and design of new composite materials, smart and multifunctional materials and metamaterials as part of the strategic research line "Digital World, Industry, Space, Defense". Specifically, (1) In-silico/In-Lab active metamaterial characterisation, (2) high-fidelity computational modelling and (3) the use of 4D printing, have been identified as three key enablers to address some of the six Grand Challenges faced by Europe, as part of the new Industry 4.0 initiative. Specifically, novel soft-active metamaterials are seen as being instrumental in leading the creation of groundbreaking advancements in five highly transformative sectors: (1) Soft Robotics, (2) Energy harvesters, (3) Data-Science Engineering, (4) Optimum soft-active control for Zero Emissions, (5) Accurate and fast biomedical Drug Delivery. Innovation in these areas of science and technology cannot be individually achieved through either trial-and-error experimental testing, due to the challenging environments in which some of these phenomena take place (e.g., soft robots operating in radioactive environments or novel drug delivery systems embedded within a human body) or via traditional computational techniques (not equipped to take advantage of Big Data). The seminal "2015-2035 NASA Roadmap Report" states that innovation can only be accomplished through 3D prototyped data-driven digital twins capable of generating high-fidelity simulations based on realistic in-lab multi-physics laws. Thus, it is imperative to exploit (1) high-fidelity computational models assisted by 3D printing (and in-lab experiments) and (2) cutting-edge data-driven assimilation techniques, in order to develop new disruptive soft-active metamaterials and methods which can ultimately drive the creation of innovative solutions in the above five sectors. Meeting these challenges, POTENTIAL puts together transversal tools such as Machine Learning and Data Science, optimization, uncertainty quantification and advanced multiphysic & multiscale simulation into an open-source virtual environment that truly aid the design of smart composites and structures based on multifunctional soft metamaterials. The project POTENTIAL is ambitious due to its multi-disciplinarity across different scientific fields, namely, (1) Continuum mechanics, (2) Computational Mechanics, (3) Applied Mathematics, (4) Material discovery and applications.

Due to the cross-cutting nature of POTENTIAL, scientific impact can be expected in, at least, two of the strategic research lines of the strategic sector Digital world, Industry, Space, Defense identified in EECTI 2021-207, namely,

(1) Modelling, mathematical analysis and new mathematical solutions for science and technology,

(2) New materials and manufacturing techniques.

INSTITUTION

MultiSimO Lab (UPCT)

Jesús Martínez-Frutos

Rogelio Ortigosa Martínez

Francisco Periago

Mathieu Kessler

INSTITUTION

CIMNE
(UPC)

Javier Bonet

Alberto García Gonzalez

External institution

Zienkiewicz Institute
(Swansea University)

Antonio J. Gil

External institution

Cyber-Physical Simulation
(TU Darmstadt)

Oliver Weeger

Subproject 1: "Data driven computational engineering for smart soft actuation" (PID2022-141957OB-C22)

The sub-project Data driven computational engineering for smart soft actuation within the coordinated project POTENTIAL focuses on the development of a computational environment embedding cutting-edge topology optimisation techniques, machine learning, dimensionality reduction and reduced order modelling techniques, efficient and robust discretisation techniques and solvers capable of addressing the open problem of multi-physics driven soft actuation. Soft multi-functional materials are endowed by improved flexibility, adaptability, and reconfigurability, which are intrinsic to living systems. These properties make them particularly promising for different applications, including soft electronics, surgery, drug delivery, artificial organs, prosthesis, and soft robotics. The additional degree of freedom for soft actuator devices can be provided through the use of intelligent materials, which are able to change their structure, macroscopic properties, and shape under the influence of external signals (electric, magnetic, thermal, etc. stimuli). However, the controllability of soft multi-functional materials, i.e., the a priori determination of the required external inputs (voltage/magnetic field) triggering the required flexible motions, is hampered by their underlying complex multi-scale and multi-physics dynamics. A purely experimental approach, based on trial-and-error decisions, is not the most effective mechanism for this type of applications. The objective of our project is to overcome this barrier adopting an alternative computer-based paradigm, supported with experimental validation, and embedding cutting-edge dimensionality algorithms. The latter will finally permit to pave the way for the crystallisation of real-time computer-based prediction of the dynamics of multi-functional materials.


Subproject 2: “Multiphysics-Informed Design of Tunable Smart Materials” (PID2022-141957OA-C22)

The sub-project Multiphysics-informed design of tuneable smart materials within the coordinated project POTENTIAL focuses on the development of a computational environment embedding cutting-edge topology optimisation techniques, machine learning, dimensionality reduction and reduced order modelling techniques, efficient and robust discretisation techniques and solvers capable of addressing the open problem of design of tuneable smart metamaterials. Metamaterials are architected composites carefully designed to show superior properties to those available in nature, such as ultrahigh stiffness and strength-to-weight ratio, or unusual properties, such as a negative Poissons ratio, a negative coefficient of thermal expansion. Recent metamaterial designs have been able to exhibit programmable shape transformations and tuneable mechanical properties. Others have shown excellent energy absorption properties. Additionally, biomechanical applications have in parallel emerged, and new metamaterials have been designed as scaffolds to promote bone cell formation.

This project aims to take a metamaterial approach towards the design of soft machines. However, this inevitably leads to an increase in the number of degrees of freedom in deformation and the available geometrical parameters. Although this makes design considerably more challenging, it also offers exciting opportunities to embed robots with sensing, actuating, and interactive functionalities, not accessible through conventional approaches. The challenge however, inherent to the across scales interactions (from micro/meso to macro) and the multi-physic nature of multi-functional materials, prevents also here to advocate for just a unidisciplinary approach to unravel the immense potential of stimuli-responsive metamaterials in soft actuation applications. The objective of our project is to advocate for a multi-disciplinary approach, requiring the development of a computer platform endowed with cutting-edge topology optimisation and machine learning techniques, counting in parallel on invaluable experimental validation in the physical lab. Our vision is that only through this paradigm, it is possible to produce a striding impact in the new generation of multi-functional smart materials.

 

UPC Repository

"Digital repository of journal articles, whose objective is to organize, archive, present, and disseminate in open access mode the intellectual production resulting from  POTENTIAL project."

UPCT Repository

"Digital repository of journal articles, whose objective is to organize, archive, present, and disseminate in open access mode the intellectual production resulting from  POTENTIAL project."

Zenodo Community

"Digital repository whose objective is to organize, archive, present, and disseminate in open access mode the Insilico and experimental data resulting from  POTENTIAL project."


Github Repository

"Digital repository whose objective is to organize, archive, present, and disseminate in open access mode the software resulting from  POTENTIAL project."



Publications

2024

A computational framework for large strain electromechanics of electro-visco-hyperelastic beams

Firouzi, Nasser; Rabczuk, Timon; Bonet, Javier; Żur, Krzysztof Kamil

A computational framework for large strain electromechanics of electro-visco-hyperelastic beams Journal Article

In: Computer Methods in Applied Mechanics and Engineering, vol. 426, 2024, ISSN: 0045-7825.

Links | BibTeX

Generalised tangent stabilised nonlinear elasticity: A powerful framework for controlling material and geometric instabilities

Poya, Roman; Ortigosa, Rogelio; Gil, Antonio J.

Generalised tangent stabilised nonlinear elasticity: A powerful framework for controlling material and geometric instabilities Journal Article Forthcoming

In: International Journal for Numerical Methods in Engineering, Forthcoming.

BibTeX

Shape-programming in hyperelasticity through differential growth

Ortigosa, Rogelio; Martínez-Frutos, Jesús; Mora-Corral, Carlos; Pedregal, Pablo; Periago, Francisco

Shape-programming in hyperelasticity through differential growth Journal Article

In: Applied Mathematics and Optimization, vol. 89, no. 49, 2024, ISSN: 1432-0606.

Abstract | Links | BibTeX

Nonlinear electro-elastic finite element analysis with neural network constitutive models

Klein, Dominik; Ortigosa, Rogelio; Martínez-Frutos, Jesús; Weeger, Oliver

Nonlinear electro-elastic finite element analysis with neural network constitutive models Journal Article

In: Computer Methods in Applied Mechanics and Engineering, vol. 425, 2024, ISBN: 1879-2138.

Abstract | Links | BibTeX

Neural networks meet hyperelasticity: On limits of polyconvexity

Klein, Dominik; Ortigosa, Rogelio; Martínez-Frutos, Jesús; Weeger, Oliver

Neural networks meet hyperelasticity: On limits of polyconvexity Journal Article Forthcoming

In: Journal of the Mechanics and Physics of Solids, Forthcoming.

BibTeX

Probability-of-failure-based optimization for Random pdes through concentration-of-measure Inequalities

Ortigosa, Rogelio; Martinez-Frutos, Jesus; Periago, Francisco

Probability-of-failure-based optimization for Random pdes through concentration-of-measure Inequalities Journal Article Forthcoming

In: ESAIM: Control, Optimisation and Calculus of Variations, Forthcoming.

Abstract | Links | BibTeX

Learning nonlinear constitutive models in finite strain electromechanics with Gaussian process predictors

Pérez-Escolar, Alberto; Martinez-Frutos, Jesus; Ortigosa, Rogelio; Ellmer, Nathan; Gil, Antonio J.

Learning nonlinear constitutive models in finite strain electromechanics with Gaussian process predictors Journal Article

In: Computational Mechanics, 2024, ISBN: 1432-0924.

Abstract | Links | BibTeX

Gradient enhanced gaussian process regression for constitutive modelling in finite strain hyperelasticity

Ellmer, Nathan; Ortigosa, Rogelio; Martinez-Frutos, Jesus; Gil, Antonio J.

Gradient enhanced gaussian process regression for constitutive modelling in finite strain hyperelasticity Journal Article

In: Computer Methods in Applied Mechanics and Engineering, vol. 418, iss. PART B, pp. 116547, 2024, ISBN: 1879-2138.

Abstract | Links | BibTeX

 

 

Software

RVEs  package

5/5

The RVEs package serves as a versatile tool for mesh generation, leveraging the Gmsh-Julia-API for seamless integration. It provides a robust framework for automating mesh generation for custom model types, enabling users to automate the creation of intricate structures such as representative volume elements (RVEs). RVEs simplifies the mesh modeling process into several key stages:

  1. Geometry Setup: Define geometric shapes using fundamental entities and boolean operations.
  2. Integration with Gmsh: Incorporate shapes into Gmsh, execute boolean operations, and define physical groups.
  3. Mesh Generation: Generate meshes with customizable refinement fields.
  4. Export and Visualization: Export meshes into various formats and visualize the results.

MIMOSA  package

5/5

This is an application repository with a collection of drivers for the simulation of nonlinear Thermo-Electro-Magneto-Mehcanical problems. It is based on Gridap, a package for grid-based approximation of PDEs with Finite Element.